Has Joe Biden Improved Democratic Performance Among 65+ Old Voters So Far?
By Alexander Agadjanian
The relationship between voter age and candidate preference has been a keen point of interest throughout the 2020 election season. The Democratic primary saw age as a central cleavage between supporters of Joe Biden and Bernie Sanders. Voting intentions by older and younger voters in trial heat polls carried big implications for the general election. More recently, after Biden won the nomination, signs of Democratic gains among older voters—as much as 15 points—have emerged
But how certain of a trend is this? With vote switching on a long-term decline, any sizable shifts in subgroup voting behavior are reasonably met with skepticism. Other data on Democratic vote shares over time does not fit so well with a Biden-led improvement among older voters. Aside from the conflicting evidence, the comparison between older aged voters’ preferences from 2016 to 2020 can be tricky. Two things in particular are worth considering when comparing cross-sectional surveys from different points in time.
Comparing Cross-Sectional Polls over Time
For one, the approaches that pollsters take can differ substantially in terms of sampling strategies, weighting, and other aspects, many of which could be unknown. When evaluating an outcome like Democratic vote margin among 65+ old voters over time, holding pollster constant—i.e., checking data for the same pollster’s surveys over time—makes these methodological idiosyncrasies less of a problem and thus creates a cleaner comparison.
A second concern is that changes in subgroup voting behavior reflect something not specific to that subgroup, but instead broader trends that impact all groups. For example, given that Biden is doing better overall than Hillary Clinton did in general election polls, perhaps 65+ gains result from the national environment moving Democratic more widely. To address this, after making sure to compare polls over time within the same survey houses, it’s crucial to check how these within-survey changes among 65+ compare to within-survey changes for all respondents.
Holding Pollster Constant
I recently collected 2016 and 2020 polling result data that would allow for this type of comparison. First, I searched for prominent pollsters that had both recently polled the 2020 election and also conducted a survey prior to the 2016 election. If some pollsters had multiple surveys during each time frame, I recorded only the polls closest to each year’s Election Day. After removing pollsters without publicly available subgroup results, I was left with seven pollsters that had age 65+ and overall Democratic margin results in 2016 and 2020. (In one case—Pew ATP 2016—I used raw data.) These results appear below:
Democratic Vote Margins among Older Respondents
The top row of results for each poll focuses on older respondents (usually 65+ in age, and in one case 55+) while the bottom set covers results for all respondents as the national environment comparison. Key statistics—Biden’s and Clinton’s polling margin over Donald Trump, as well as Biden’s improvement on Clinton’s margin—appear on the right-hand side. I also include some additional relevant information, such as mode of interview, the population it samples (RV = registered voters, LV = likely voters), and the range of interview dates.
As the top right column “Dem Gain” shows, the degree of improvement in Biden’s performance among older voters compared to Clinton’s performance varies across pollsters. But each difference is in the positive direction, and they average out to a 9.6 point shift in the Democratic direction from 2016 to 2020. For example, Democratic older voter performance posts the worst marks in YouGov polls. Yet even there, whereas Clinton’s deficit was 20 points right before the 2016 election, it’s currently only eight points for Biden.
Is this Just a National Shift?
So, even when you look within the same pollsters, Biden has so far improved on Clinton's general election performance among older voters. However, what if this just reflects Biden doing better overall in 2020 polls, and thus his gains are not so specific to older aged individuals? The bottom row of results for each poll above addresses this point, providing margins among all voters in the same polls used for older voters.
If Democratic gains move in tandem across older vs. all respondents, then it’s less notable for Biden to improve among older voters. The table shows some traces of this story. For example, in CNN polls, Biden improves by eight points among 65+ and by six points among all voters. But this pattern is hardly consistent across pollsters—for the most part, these gains do not move together. Overall, Democratic gains average out to about 0 (compared to nearly a 10 point gain among older voters). In other words, this comparison to national margins reveals older voter gains as more unique than a product of broader national improvement.
Zeroing in on YouGov
It’s also worth taking a closer look at YouGov polls. Earlier evidence seemed to suggest much weaker 65+ gains there, Democrats generally perform worst among 65+ voters in these polls, and with so many surveys in both 2016 and 2020, perhaps results I show are sensitive to which poll I classify as most recent. I expand my coverage to the three most recent polls I found for 2020 and the three polls closest to 2016 Election Day, and show results below:
Among YouGov Polls Only
For this table, the top set of results is for 2020 and the bottom set is for 2016. Each set includes 65+ and national margins for Biden and Clinton, as well as the Democratic deficit among 65+ voters relative to its national margin. Biden continues to underperform among older voters compared to the entire electorate, but this deficit (an average of -13 points) is much less severe—about twice as small—than Clinton’s deficit in 2016 (-26 points).
Caveats and Conclusions
These results don’t come without caveats. Pollsters may have changed minor aspects of their methodologies from 2016 to 2020. In one case, a pollster (CNN) changed the vendor used to conduct its poll, so results there should be viewed carefully. Another issue is that 2016 polls largely used likely voter populations whereas 2020 polls use registered voter ones. It’s unclear whether population effects (going from RV to LV) differ in 2016 vs. 2020—which would undercut this analysis—but it’s worth keeping in mind (note that if there was a difference, it would show up for all voters, so comparing against national margins helps here too).
In sum, the results here provide pretty clear evidence that Biden has so far improved quite a bit on Clinton’s performance among older voters. On average, comparing the same pollsters for the 2016 and 2020 elections, Biden is running 10 points better among 65+ voters than Clinton did four years ago. That voting preferences tend to stick over the course of a campaign make these gains all the more important. They should still be tracked in the months to come, as well as get further parsed (e.g. Is this generational change? Are there racial or class elements?). But for now, Biden’s gains among old voters represent a clear positive sign for Democrats as we head into the general election campaign.
Alexander Agadjanian (@A_agadjanian) is a research associate at the MIT Election and Data Science Lab.